Re: Design of pg_stat_subscription_workers vs pgstats

On Thu, Feb 3, 2022 at 2:35 PM Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> wrote:
>
> On Thu, Feb 3, 2022 at 1:48 PM David G. Johnston
> <david(dot)g(dot)johnston(at)gmail(dot)com> wrote:
> >
> > On Wednesday, February 2, 2022, Masahiko Sawada <sawada(dot)mshk(at)gmail(dot)com> wrote:
> >>
> >> and have other error
> >> information in pg_stat_subscription_workers view.
> >
> >
> > What benefit is there to keeping the existing collector-based pg_stat_subscripiton_workers view? If we re-write it using shmem IPC then we might as well put everything there and forego using a catalog. Then it behaves in a similar manner to pg_stat_activity but for logical replication workers.
>
> Yes, but if we use shmem IPC, we need to allocate shared memory for
> them based on the number of subscriptions, not logical replication
> workers (i.e., max_logical_replication_workers). So we cannot estimate
> memory in the beginning. Also, IIUC the number of subscriptions that
> are concurrently working is limited by max_replication_slots (see
> ReplicationStateCtl) but I think we need to remember the state of
> disabled subscriptions too.

Correction; the replication state remains even after the subscription
is disabled, and it's removed only when the subscription is dropped.
Therefore, the number of subscriptions that can be active in the
database cluster is effectively limited by max_replication_slots.

Regards,

--
Masahiko Sawada
EDB: https://www.enterprisedb.com/